Error before Linux starts (both bootlets and u-boot)

Started by daparix, August 29, 2013, 06:59:00 PM

Previous topic - Next topic



I am trying to connect an LCD to the iMX233, so I have done some patches to the kernel and did a wrap board following some instructions from Claude.

The board works if I use original kernel from Claude. So it looks like the HW is ok.

If I build a kernel and use either bootlets or u-boot, with the board UNPLUGGED, then it boots, and I get the framebuffer.

If I plug the board and boot same microSD, then I get an early error (board reboots continuously). With bootlets it is:

PowerPrep start initialize power...

Configured for 5v only power source.            Battery powered operation disab.


PowerPrep start initialize power...

Configured for 5v only power source.            Battery powered operation disab.

I have seen in the forum that the "r14_HTLLLL" happens also when bootlets are built with wrong compiler (it says "Invalid Instruction" too).

I would like to have some hint about where is the failing code (I guess it is in the bootlets source, but did not find any string like that). I guess it is some kind of unexpected interrupt, but I can not find the place to look at the code.

Anyone can tell me where the error can be located so I can debug? The only option I see is to put a lot of traces in the bootlets code so I can see where it stops.

thank you



i put some debugging info on all bootlets modules.

problem was at the first one, "power_prep". I was using a 5V live source, so I selected "no-battery" in the code by defining directive NO_BATTERY_VOLTAGE_SOURCE. I attempted to undefine it and use the normal boot with battery check... and it works.

You need to connect pin 1 (BAT) in imx233 GPIO connector to GND so the bootlet works... if you leave unconnected it sometimes detects some voltage and does not boot.

...Probably I found some kind of bug in the bootlets, that does not raise unless you try to use LCD and define the "no-battery" directive. I have no real idea of the bug, but at least I can move on.

Once I finish the tests, will provide schemas and patches to connect the LCD as Claude originally proved.